API or serverKnowledge management
MemOS
Store, retrieve, and compose long-term memory for LLM applications and agents.
- Stars
- 10,207
- Forks
- 930
- Updated
- Updated Jul 14, 2026
Apache-2.0
Store, retrieve, and compose long-term memory for LLM applications and agents.
MemOS exposes one API for adding, searching, editing, and deleting agent memory, with inspectable graph structure for text, images, tool traces, and personas. Separate memory cubes can isolate or selectively combine knowledge across users, projects, and agents. Deployment options include a managed cloud API, a self-hosted service using Neo4j and Qdrant, or local Hermes and OpenClaw plugins backed by on-device SQLite.
Resource types
Use cases
Runtime
Write, draw, organize knowledge, and plan in one local-first workspace.
Add RAG, files, tools, databases, and MCP APIs to locally hosted language models.
Protocols & integrations
Capabilities
Public GitHub facts last synced Jul 14, 2026.
Backend services and hosting for web, mobile, and AI apps
Manage local notes, documents, and tasks with blocks and bidirectional links.